Predictive maintenance of sand accumulations
Unpredicted and unexpected sand accumulations in desert regions jeopardize critical infrastructure and therefore increase the risk of operation disruptions - e.g. in oil field, and the interruptions of daily operations can lead to significant revenue losses for companies in the region. Therefore, a leading NOC wanted to leverage existing internal and external data sources to predict future sand accumulations. The resulting predictive maintenance concept could lead to a significant decrease in maintenance-related costs.
To prevent or handle the risk of sand accumulations and disruptions, a lot of time and efforts need to be invested by manned patrols to help identify blocked areas manually. Additionally, much manual manpower is required to load and unload sand in non-critical areas. This can result in daily interruption of operations which often lead to significant revenue losses for the company.
We leveraged existing, operational and GIS data to predict sand accumulates based on machine-learning algorithms.
- We identified areas of highest improvement potential based on existing operational data.
- We correlated data on clearing activities for every geo-point and trained machine-learning algorithms to predict accumulations.
- We enriched our model with pixel segmentation based on satellite images.
We expect a 27% reduction of maintenance-related costs through the predictive AI-solution for accumulation identification and automation of the monitoring process developed by Siemens Advanta.
Our data-driven prediction model led to several benefits for our client:
Decrease of maintenance-related costs by 27% through manhour reduction and increase of process efficiency via forecasting automation
Improvement of health and safety for employees as well as the carbon footprint of the company
Reduced risk of disruptions of operations and therefore higher plant uptimes possible